Initial Value Problems vs. Boundary Value Problems

A differential equation on its own has infinitely many solutions. Side conditions pick one out. When all the conditions are imposed at a single value of the independent variable, the result is an initial value problem. When they are imposed at two or more separate points, it is a boundary value problem. The distinction is not cosmetic: it changes which solutions exist, whether they are unique, and how you go about finding them.

Quick Reference

	Initial Value Problem (IVP)	Boundary Value Problem (BVP)
Where conditions are given	All at the same point $x_{0}$	At two or more separate points
Typical second-order form	$y (x_{0}) = α$ , y'(x_0) = \beta	$y (a) = α$ , $y (b) = β$
Example	y'' + 4y = 0, $y (0) = 0.2$ , y'(0) = 0.1	y'' + 4y = 0, $y (0) = - 0.1$ , $y (π / 4) = 0.2$
Existence and uniqueness	Guaranteed for linear equations with continuous coefficients	Not guaranteed; may have one, none, or infinitely many solutions
Natural setting	Evolution in time from a known starting state	Steady state with conditions at spatial boundaries
Numerical approach	March forward from the initial data	Iterate until conditions at all boundaries are met

Why Side Conditions Are Needed

Often, in addition to the differential equation, the unknown function is required to satisfy certain conditions at one or more points within the interval of interest.

Recall that an $n$ th order equation has a general solution containing $n$ arbitrary constants. The equation y'' + 4y = 0, for example, has general solution

y = c_{1} \cos 2 x + c_{2} \sin 2 x,

an entire two-parameter family of curves. To single out one solution you need two more pieces of information, and where you place them determines what kind of problem you have.

Initial Value Problems

When the side conditions are specified at the same value of $x$ (or $t$ , or whatever the independent variable is), typically at the beginning of the interval where the solution is sought, they are called initial conditions. A differential equation together with initial conditions is called an initial value problem (IVP).

An initial value problem involves a differential equation along with additional conditions specified at a single point. For example, given an equation like $\frac{d y}{d x} = f (x, y)$ , an IVP also provides the value of $y$ at a particular point $x_{0}$ , such as $y (x_{0}) = y_{0}$ .

The three equations

y'' + 4y = 0, \quad y(0) = 0.2, \quad y'(0) = 0.1

constitute an initial value problem, because both conditions are specified at $x = 0$ .

The key characteristic of an IVP is that the initial conditions are given at one specific point, and the solution is typically sought on an interval starting from that point.
IVPs are common in problems where the state at a starting time is known and the objective is to predict future behavior, such as the motion of a projectile at the moment it is launched.

Solve the initial value problem y'' + 4y = 0, $y (0) = 0.2$ , y'(0) = 0.1.

Solution

The general solution is $y = c_{1} \cos 2 x + c_{2} \sin 2 x$ , with derivative

y' = -2c_1 \sin 2x + 2c_2 \cos 2x .

Apply the conditions at $x = 0$ , using $\cos 0 = 1$ and $\sin 0 = 0$ :

\begin{aligned} y(0) &= c_1 = 0.2 \\ y'(0) &= 2c_2 = 0.1 \end{aligned}

So $c_{1} = 0.2$ and $c_{2} = 0.05$ , giving the unique solution

y = 0.2 \cos 2 x + 0.05 \sin 2 x .

Notice how clean the algebra is. Because both conditions were imposed at the same point, each one determined a single constant directly, with no simultaneous system to untangle.

Boundary Value Problems

If the conditions are specified at the endpoints, or boundaries, of the interval, they are called boundary conditions. A differential equation together with boundary conditions is called a boundary value problem (BVP).

A boundary value problem involves a differential equation along with conditions specified at two or more points. For the equation \dfrac{d^2y}{dx^2} = f(x, y, y'), a BVP would specify values at two points, such as $y (a) = α$ and $y (b) = β$ , where $a$ and $b$ are the endpoints of the interval of interest. An example of such a problem is

y'' + 4y = 0, \quad y(0) = -0.1, \quad y(\pi/4) = 0.2

BVPs are characterized by the requirement to satisfy conditions at multiple boundaries, which can make them more complex to solve than IVPs.
Such problems are typical in steady-state phenomena where conditions are known at spatial boundaries, like the temperature distribution along a rod that is heated at both ends.

Solve the boundary value problem y'' + 4y = 0, $y (0) = - 0.1$ , $y (π / 4) = 0.2$ .

Solution

Again start from $y = c_{1} \cos 2 x + c_{2} \sin 2 x$ .

The first condition at $x = 0$ gives $c_{1} = - 0.1$ .

The second condition is at $x = π / 4$ , where $2 x = π / 2$ , so $\cos (π / 2) = 0$ and $\sin (π / 2) = 1$ :

y (π / 4) = c_{1} \cdot 0 + c_{2} \cdot 1 = c_{2} = 0.2 .

The solution is

y = - 0.1 \cos 2 x + 0.2 \sin 2 x .

This particular BVP behaved well, but that was luck. The next example shows what usually makes boundary value problems harder.

Existence and Uniqueness Behave Differently

This is the deepest difference between the two kinds of problem, and it is worth seeing concretely.

For a linear equation with continuous coefficients, an initial value problem always has exactly one solution. A boundary value problem may have exactly one solution, no solution at all, or infinitely many.

Consider y'' + 4y = 0 on $[0, π / 2]$ with the boundary conditions below. Determine how many solutions each problem has.

$y (0) = 0$ , $y (π / 2) = 0$
$y (0) = 0$ , $y (π / 2) = 1$

Solution

Start from $y = c_{1} \cos 2 x + c_{2} \sin 2 x$ . At $x = π / 2$ we have $2 x = π$ , so $\cos π = - 1$ and $\sin π = 0$ . Therefore

y (0) = c_{1}, y (π / 2) = - c_{1} .

Case 1. The conditions read $c_{1} = 0$ and $- c_{1} = 0$ . Both are satisfied by $c_{1} = 0$ , and $c_{2}$ is left completely free. Every function

y = c_{2} \sin 2 x

solves the problem, so there are infinitely many solutions.

Case 2. The conditions read $c_{1} = 0$ and $- c_{1} = 1$ , which demand $c_{1} = 0$ and $c_{1} = - 1$ at the same time. That is impossible, so there is no solution.

Compare this with the IVP y'' + 4y = 0, $y (0) = α$ , y'(0) = \beta, which has exactly one solution for every choice of $α$ and $β$ . The boundary value problem is far more delicate, because the two conditions must be met by the same pair of constants at two different places, and the solution's behavior in between is not free to adjust.

The values of a parameter for which a BVP has nontrivial solutions are called eigenvalues, and the corresponding solutions are eigenfunctions. Case 1 above is exactly this phenomenon. It is the foundation of the boundary value problem chapter and of Fourier series methods for partial differential equations.

Why the Solution Methods Differ

The methods for solving IVPs and BVPs can differ significantly, primarily due to the nature of the conditions that must be satisfied.

Numerical methods for IVPs often proceed forward from the known initial condition. Since the state at $x_{0}$ is fully specified, a stepping method such as Euler's method or a Runge-Kutta method can advance the solution one small step at a time, using only information already computed.

BVPs allow no such march, because the data at the far boundary is not available at the start. They require iterative techniques that adjust the solution until it satisfies the conditions at all specified boundaries. The shooting method, for example, guesses the missing initial slope, solves the resulting IVP, checks the value reached at the far boundary, and revises the guess. Other approaches discretize the entire interval at once and solve a large system of algebraic equations.

Exercises

Classify each problem as an IVP or a BVP.

y'' - y = 0, $y (1) = 2$ , y'(1) = -3
y'' + y = 0, $y (0) = 1$ , $y (π) = 1$
y''' = x, y(0) = y'(0) = y''(0) = 0

Solution

IVP. Both conditions are at $x = 1$ .
BVP. The conditions are at the two distinct points $x = 0$ and $x = π$ .
IVP. All three conditions are at $x = 0$ , matching the third order of the equation.

Given that $y = c_{1} e^{x} + c_{2} e^{- x}$ is the general solution of y'' - y = 0, solve the IVP with $y (0) = 3$ and y'(0) = 1.

Solution

Since y' = c_1 e^{x} - c_2 e^{-x}, the conditions at $x = 0$ give

\begin{aligned} c_1 + c_2 &= 3 \\ c_1 - c_2 &= 1 \end{aligned}

Adding gives $2 c_{1} = 4$ , so $c_{1} = 2$ and then $c_{2} = 1$ . The solution is

y = 2 e^{x} + e^{- x} .

Show that the boundary value problem y'' + y = 0, $y (0) = 0$ , $y (π) = 5$ has no solution.

Solution

The general solution is $y = c_{1} \cos x + c_{2} \sin x$ . At $x = 0$ , $y (0) = c_{1} = 0$ . At $x = π$ , $\cos π = - 1$ and $\sin π = 0$ , so

y (π) = - c_{1} = 0 .

Every candidate solution is forced to vanish at $x = π$ , so the requirement $y (π) = 5$ can never be met. The problem has no solution.

Note that the corresponding IVP with conditions $y (0) = 0$ and y'(0) = 5 has the perfectly good solution $y = 5 \sin x$ . Nothing is wrong with the equation; the difficulty comes entirely from where the conditions were placed.

Frequently Asked Questions

What is the difference between an initial condition and a boundary condition?

Only the location. Both prescribe values of the solution or its derivatives. Initial conditions are all imposed at a single point, usually the start of the interval. Boundary conditions are imposed at two or more distinct points, usually the endpoints of the interval. The equation itself does not change.

How many conditions does an

n

th order equation need?

Exactly $n$ , since the general solution carries $n$ arbitrary constants and each condition fixes one of them. A second-order equation needs two, a third-order equation needs three. Supplying fewer leaves a family of solutions; supplying more usually makes the problem unsolvable.

Why can a boundary value problem have no solution when the same equation as an IVP always works?

An IVP fixes the solution's value and slope at one point, and the equation then determines everything that follows. Nothing can conflict. A BVP instead demands that the solution, having left the first boundary, arrive at the second boundary with a prescribed value. The equation controls what happens in between, so the arrival value is not free. If the required arrival value is unreachable, no solution exists.

Must an initial condition be given at

t = 0

No. The name is traditional but the point $x_{0}$ can be anywhere. The conditions $y (3) = 1$ , y'(3) = -2 form a perfectly ordinary initial value problem. What matters is that all conditions share the same location, not that the location is zero.

Which type appears more often in applications?

Both are common, and they tend to split along time versus space. Problems that evolve in time from a known starting state are naturally IVPs: a launched projectile, a decaying isotope, a circuit switched on at $t = 0$ . Problems describing a steady state across a region are naturally BVPs: the deflection of a beam supported at both ends, the temperature profile in a rod held at fixed temperatures at each end, the shape of a hanging cable.

Can a problem mix derivative and value conditions at a boundary?

Yes, and this is standard. A condition prescribing the value $y (a) = α$ is called a Dirichlet condition, one prescribing the derivative y'(a) = \alpha is a Neumann condition, and a combination such as y(a) + 2y'(a) = 0 is a Robin or mixed condition. All three appear in the boundary value problem chapter.

Initial Value Problems vs. Boundary Value Problems

Quick Reference

Why Side Conditions Are Needed

Initial Value Problems

Boundary Value Problems

Existence and Uniqueness Behave Differently

Why the Solution Methods Differ

Exercises

Frequently Asked Questions

Quick Links

Social Media